patterns of decadal climate variability and their impact on rainfall and the biosphere peter g....
TRANSCRIPT
Patterns of Decadal Climate Variability and their Impact on
Rainfall and the Biosphere
Peter G. BainesDept of Civil and Environmental
Engineering, University of Melbourne Australia
The principal patterns of variation of the observed climate over the past 100+ years, and the dynamics behind them, are coming into focus.
Here I will describe these patterns, and then say something about how they affect rainfall, and also some connections between rainfall and the biosphere.
Of all the variables that one may use to describe climate, the most important is (probably) Sea Surface Temperature (SST), because:
1.It gives a measure of heat storage in the ocean.
2.It has a controlling effect on surface winds and pressure, over the ocean.
3.It has a controlling effect on surface humidity.
The collection and interpretation of observations over the past 200 years or so now enable the estimation of global observations back to about 1850.
This has mainly been done by two separate groups: the Hadley Centre at the UKMO, and Smith, Reynolds & c at NOAA /US, with ongoing upgrades.
There are many differences between these two data sets. These differences mainly arise from the manner in which the observations are interpolated into regions where there are none.
References: HadISST1 (Rayner et al. 2003 ) erSSTv3 (Smith et al. 2008)
Here, I have used data from each set from 1900 to 2009, taking annual means over years that begin in June and end the following May. These are termed “ENSO-years” , or E-years for short. Five-year running means have then been taken over these E-years. This covers time scales longer than ENSO.
An EOF analysis is then performed on the resulting data. This gives a breakdown into patterns that may be associated with recognisable physical processes.
This procedure is most effective if the values of S(k,k) for k = 1,2,3.. decrease rapidly, so that most of the rest of them are small.
We then have a succinct description of the data.
EOFs are mathematical artifices that are effectively efficient descriptions of the data, and any particular one may or may not be a dynamical entity by itself.
Via the Singular Value Decomposition Theorem (SVD), a data matrix Rij to be expressed in the form
n
k
Tij kjVkiUkkSR
1
,),(),( assuming m > n
where the SUV denote Empirical Orthogonal Functions (EOFs), S(k,k)2 denotes the variance of the k th EOF, andU and V are orthogonal matrices.
Spatial pattern Time series
East Longitude
Latitu
de
Mean of 5Eyr means of HadISST jun1900-may2009
50 100 150 200 250 300 350
-30
-20
-10
0
10
20
30
40
50
60
5
10
15
20
25
Mean Sea Surface Temperature, Hadley Centre HadISST1 data
Data are: 5-year running means of HadISST1 and Smith-Reynolds erSSTv3, from June1900-May 2009.5-year running means cover variance on time scales longer than ENSO
East Longitude
Latit
ude
EOF1 of HadISST1, 5Eyrmean, June1900-May2009
50 100 150 200 250 300 350
-60
-40
-20
0
20
40
60
-1
0
1
2
3
4
5
6
7
8
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
Central E-years from 1902
V(:
,1)
Time series for EOF1: blue-HadISST, red-erSST. solid-70S, dashed-40S
HadISST1
erSSTv3
Southern boundary at 70S - solid curves, 40S - dashed
EOF1 - GW
Comparisons – 70N-70SLeft – HadISSTRight – erSST
EOF1
The Global Warming PatternVariance 51.5 & 57.5%
(-)EOF1 of erSSTv3, 5E-year running mean, June1900-May2009
East Longitude
Latit
ude
0 50 100 150 200 250 300 350
-60
-40
-20
0
20
40
60
-2
-1
0
1
2
3
4
5
6
Properties of the Global Warming Pattern:
1. Contains the secular trend in the data, similar to global mean temperature (IPCC)
2. Contains more than half the total variance in the data
3. The spatial pattern is approximately uniform, and mainly due to radiative processes, influenced by increasing greenhouse gases and aerosols.
4. There are some small exceptions to this uniform heating: The region near Antarctica, implying little change there A small region in the northern North Atlantic, which has cooled. - the latter may be attributed to a small slowing in the “Atlantic conveyor” (according to numerical model studies).
5. “Wet regions get wetter, dry regions get dryer” (Held & Soden2006)
East Longitude
Latit
ude
EOF2 of HadISST1, 5Eyrmean, June1900-May2009
50 100 150 200 250 300 350
-60
-40
-20
0
20
40
60
-4
-3
-2
-1
0
1
2
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
Central E-years from 1902
V(:
,2)
Time series for EOF2: blue-HadISST, red-erSST. solid-70S, dashed-40S
HadISST1 erSSTv3
Southern boundary at 70S - solid curves, 40S - dashed
EOF2 - PDO
Comparisons – 70N-70SLeft – HadISSTRight – erSST
EOF2
The Pacific Decadal Oscillation PDO/IPO PatternVariance: 14 & 11%
East Longitude
Latit
ude
EOF2 of erSSTv3, 5E-year running mean, June1900-May2009
0 50 100 150 200 250 300 350
-60
-40
-20
0
20
40
60
-4
-3
-2
-1
0
1
2
3
East Longitude
Latit
ude
EOF3 of HadISST1, 5Eyrmean, June1900-May2009
50 100 150 200 250 300 350
-60
-40
-20
0
20
40
60
-2
-1
0
1
2
3
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010-0.25
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
Central E-years from 1902
V(:,3)
Time series for EOF3: blue-HadISST, red-erSST. solid-70S, dashed-40S
Southern boundary at 70S - solid curves 40S - dashed
EOF3 - AMO
HadISST1
erSSTv3
Comparisons – 70N-70SLeft – HadISSTRight – erSST
EOF3
Atlantic Meridional Oscillation AMO PatternVariance: 7.5 & 6.8%
East Longitude
Latit
ude
EOF3 of erSSTv3, 5E-year running mean, June1900-May2009
0 50 100 150 200 250 300 350
-60
-40
-20
0
20
40
60
-2
-1
0
1
2
3
East Longitude
Latit
ude
EOF4 of HadISST1, 5Eyrmean, June1900-May2009
50 100 150 200 250 300 350
-60
-40
-20
0
20
40
60
-3
-2
-1
0
1
2
3
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010-0.25
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
HadISST1
erSSTv3
Time series for EOF4: blue-HadISST, red-erSST. solid-70S, dashed-40S
Southern boundary at 70S - solid curves, 40S - dashed
Central E-years from 1902
V(:
,4)
EOF4 - PGO
Comparisons – 70N-70SLeft – HadISSTRight – erSSTEOF4
The Pacific Gyre Oscillation PGO PatternVariance: 5.1 & 4%These first four EOFs contain ~ 80% of variance for each data set.
EOF4 of erSSTv3, 5E-year running mean, June1900-May2009
East Longitude
Latit
ude
0 50 100 150 200 250 300 350
-60
-40
-20
0
20
40
60
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
The above describes 54.8% and 51.3% of the variance of the remainder of each of these two data sets after EOF1 has been removed.
However, we may do better with complex EOFs.
This involves taking the Hilbert transform of the time series T(r,t) at each spatial grid point of the data, and regarding this as the imaginary part of the time series, giving the complex time series at each position r : T(r,t) + i H(T(r,t)) One then takes the SVD of this complex matrix to obtain the complex CEOFs, again taking the real part of the results for real values.
n
k
TCCCij kjVkiUkkSR
1
,),(),(
where Uc and Vc are now complex. The following calculations are for HadISST1
Complex EOF1 of HadISST1(-EOF1),
5Eyrmeans, 1900-2009
(2nd EOF overall)
The PDO/IPO36.7% of total variance
The total EOF is:
realV(:,1).CEOF1re + imag(V:,1).CEOF1im
East longitude
Latit
ude
CEOF1re-HadISST1900(-EOF1)-5Eyr-160W
50 100 150 200 250 300 350
-60
-40
-20
0
20
40
60
-5
-4
-3
-2
-1
0
1
2
3
East longitude
Latit
ude
(-)CEOF1im-HadISST1900(-EOF1)-5Eyr-160W
50 100 150 200 250 300 350
-60
-40
-20
0
20
40
60
-6
-4
-2
0
2
4
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Central years of 5 Eyear means
real(V
(:,1
)),(
-)im
ag(V
(:,1
))
Time series of real and imag parts of CEOF1 of HadISST1(-EOF1), 5Eyearmeans
realV(:,1)
(-)imagV(:,1)
The Complex CEOF1 Cycle
(was EOF2)
East longitude
Latit
ude
CEOF1re-HadISST1900(-EOF1)-5Eyr-160W
50 100 150 200 250 300 350
-60
-40
-20
0
20
40
60
-5
-4
-3
-2
-1
0
1
2
3
East longitude
Latit
ude
(-)CEOF1im-HadISST1900(-EOF1)-5Eyr-160W
50 100 150 200 250 300 350
-60
-40
-20
0
20
40
60
-6
-4
-2
0
2
4
East longitude
Latit
ude
CEOF1imHadISST1900(-EOF1)-5Eyr-160W
50 100 150 200 250 300 350
-60
-40
-20
0
20
40
60
-4
-2
0
2
4
6
East longitude
Latit
ude
(-)CEOF1reHadISST1900(-EOF1)-5Eyr-160W
50 100 150 200 250 300 350
-60
-40
-20
0
20
40
60
-3
-2
-1
0
1
2
3
4
5
-0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15The complex time series for CEOF1 - the PDO/IPO, HadISST1 data, 5Eyrmeans
1902
2006
1912
1922
1932
1942
1962
1982
1952
1972
1992
2002
Real Part Re(Vc(:,1))
Imag
Par
t (I
m(-
)Vc(
:,1)
)
Properties of the PDO/IPO:
1. The cycle is focussed on the eastern equatorial Pacific, with northern and southern mid- latitudes in anti-phase.
2. Possibly forced by (relatively) high frequency ENSO events
3. Responsible for low-frequency changes in the equatorial Pacific and associated ENSO characteristics
4. A major regime shift occurred in the mid-1970s (Zhang&c1997)
Despite a lot of attention, still not well understood
Complex EOF2 of HadISST1(-EOF1),
5Eyrmeans, 1900-2009
(3rd EOF overall)
The AMO17.7% of total variance
The total EOF is:
realV(:,2).CEOF2re + imag(V:,2).CEOF2im
East longitude
Latit
ude
CEOF2im-HadISST1900(-EOF1)-5Eyr-160W
50 100 150 200 250 300 350
-60
-40
-20
0
20
40
60
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
3
East longitude
Latit
ude
CEOF2re-HadISST1900(-EOF1)-5Eyr-160W
50 100 150 200 250 300 350
-60
-40
-20
0
20
40
60
-2
-1
0
1
2
3
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Central years of 5 Eyear means
real(V
c(:
,2))
,im
ag(V
c(:
,2))
Time series of real and imag parts of CEOF2 of HadISST1(-EOF1), 5Eyearmeans
realV(:,2)
imagV(:,2)
Properties of the AMO:
1. - associated with the “Atlantic conveyor”
2. SST signature is the “Hemispheric pattern”, with the largest signal in the northern North Atlantic
3. Modelling studies (HadCM3) show an inbuilt oscillation with decadal time scales.
4. A major regime shift occurred in the late 1960s (Baines & Folland 2005).
Complex EOF3 of HadISST1(-EOF1),
5Eyrmeans, 1900-2009
(4th EOF overall)
The PGO11.5% of total variance
CEOFs 1,2+3 contain 66% of variance
The total EOF is:
realV(:,3).CEOF3re + imag(V:,3).CEOF3im 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010
-0.1
-0.05
0
0.05
0.1
0.15
Time series of real and imag parts of CEOF3 of HadISST1(-EOF1), 5Eyearmeans
imagVc(:,3)
realVc(:,3)
Central years of 5 Eyear means
real(V
c(:
,3))
,im
ag(V
c(:
,3))
East longitude
Latit
ude
CEOF3re-HadISST1900(-EOF1)-5Eyr-160W
50 100 150 200 250 300 350
-60
-40
-20
0
20
40
60
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
East longitude
Latit
ude
CEOF3im-HadISST1900(-EOF1)-5Eyr-160W
50 100 150 200 250 300 350
-60
-40
-20
0
20
40
60
-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
-0.1 -0.05 0 0.05 0.1 0.15
-0.1
-0.05
0
0.05
0.1
0.15The complex time series for CEOF3 - the PGO, HadISST1 data, 5Eyrmeans
Real Part (ReVc(:,3))
Imag
Par
t (Im
Vc(
:,3))
19021912
1922
1932
1942
1952
1962
1972
1982
1992
2006
2002
Properties of the PGO:
1. The focus is on mid-latitudes in the North and South Pacific
2. It involves variations in the strength of the oceanic gyre, with apparently regular periodicity of ~ 35 years.
3. The mechanism for oscillations as identified by models (Latif & Barnett (1994/6)) involves thermal forcing of the atmosphere with subsequent slow evolution of the ocean.
Rainfall
Longitude
Latitu
de
Mean rainfall over 104 5Eyear means, UDelaware data, mm/day
-150 -100 -50 0 50 100 150-60
-40
-20
0
20
40
60
5
10
15
20
25
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
Central E-years
(-)V
(:,1)
Time series for EOF1 of HadISST1
Longitude
Latitu
de
Projection of EOF1 of HadISST1 on UDel rainfall, 5Eyrmeans, mm/day
-150 -100 -50 0 50 100 150
-40
-20
0
20
40
60
-15
-10
-5
0
5
Projection of EOF1 of HadISST1 on Udelaware rainfall
Biology - fAPAR
Longitude
Latit
ude
faparMean98-05N, from faparMonNew
-150 -100 -50 0 50 100 150-60
-40
-20
0
20
40
60
80
0.1
0.2
0.3
0.4
0.5
0.6
Annual Mean
Longitude
Latitu
de
EOF1-fapar Sep97-Jun06
-150 -100 -50 0 50 100 150-60
-40
-20
0
20
40
60
80
-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
months from Sept 1997
Amplitude of EOF1-fapar, V(:,1)
EOF1 fAPAR68.6% of total variance
Longitude
Latitu
de
Annual of fAPAR for Africa over calendar years 1998-2005
-20 -10 0 10 20 30 40 50-40
-30
-20
-10
0
10
20
30
40
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1Longitude
Latitu
de
Mean Value of TRMM rainfall for Africa, 1998-2008, units: mm/hour
-20 -10 0 10 20 30 40 50
-30
-20
-10
0
10
20
30
40
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
Mean Value of TRMM RainfallUnits: mm/hour1998-2008
Mean value of fAPAR, 1998-2005
fAPAR is the fraction of Absorbed Photosynthetically Active Radiation, a measure of plant growth(Ref.: Gobron et al. 2006, J.Geophys. Res. 111, D13110)
2 4 6 8 10 12-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
June December
Months from January
V(:
,1)f
or
TR
MM
and f
AP
AR
EOF1 of annual cycle of African TRMM rainfall and fAPAR
Rainfall fAPAR
Longitude
Latit
ude
EOF1 of Annual cycle of fAPAR for Africa over years 1998-2005
-20 -10 0 10 20 30 40 50-40
-30
-20
-10
0
10
20
30
40
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Longitude
Latit
ude
EOF1 of Annual cycle of TRMM rainfall for Africa over years 1998-2008
-20 -10 0 10 20 30 40 50
-30
-20
-10
0
10
20
30
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
TRMM Rainfall fAPAR
EOF1 of the Annual cycle of TRMM Rainfall and fAPAR
Each describes 75% of the total variance of the annual cycle
1998 2000 2002 2004 2006 2008 20100
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
Months from January 1998
Rai
nfal
l (m
m/h
our)
and
fA
PA
R
Sahel monthly TRMM rainfall in mm/hour (blue) and fAPAR (red)
Rainfall
fAPAR
1998 2000 2002 2004 2006 2008 20100
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5Central Africa monthly TRMM rainfall in mm/hour (blue) and fAPAR (red)
Months from January 1998
Rai
nfal
l (m
m/h
our)
and
fA
PA
R
fAPAR
Rainfall
Longitude
Latitu
de
Mean TRMM annual rainfall over E-years June1998-May2009
-20 -10 0 10 20 30 40 50
-30
-20
-10
0
10
20
30
40
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
10 years of Sahel TRMM rainfall and fAPAR
10 years of Central African TRMM rainfall and fAPAR
Annual mean rainfall
CONCLUSIONS
1.Four patterns (GW, PDO, AMO, PGO) dominate the variability of global climate on the decadal time scale, and of Africa in particular. The PDO, AMO and PGO are oscillatory in nature, with very different dynamics and implications for rainfall and climate variability.
2.Despite the expectations that wet regions get wetter and dry regions drier from Global Warming, data over land show increased precipitation in many areas.
3.The annual cycle of plant growth (as measured by fAPAR) follows rainfall (by ~1 month), and is much less variable (in Sahel and central Africa).
EOF1 of Annual cycle of TRMM rainfall for Africa over years 1998-2008
Longitude
Latit
ude
-20 -10 0 10 20 30 40 50
-30
-20
-10
0
10
20
30
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
EOF2 of Annual cycle of TRMM rainfall for Africa over years 1998-2008
Longitude
Latit
ude
-20 -10 0 10 20 30 40 50
-30
-20
-10
0
10
20
30
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
2 4 6 8 10 12
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
Months from January
V(:
,1)
and V
(:,2
)
EOF1
EOF2
Time series of EOFs 1 and 2 for TRMM annual cycle of African rainfall
EOF1 EOF2
EOFs 1 (75%) and EOF2 (13.1%) of the Annual Cycle of TRMM rainfall for Africa
Percentages denote fractions of variance of the annual cycleNote that EOF1 is dominant